Actually this change decreases flexibility of tests. Expected output is generated. See the comment above it. When you change the bytecode, you need just uncomment these lines, run tests, copy the output, and comment out them back. The output contains numerical values of opcodes.

It is rarely they are changed without changing the meaning of opcodes. It is more common to add new opcodes and replace old opcodes with other opcodes with different semantic. This leads to changing opcode arguments and offsets, adding and removing lines.